Extending Database Space of Inada/ODMG for Very Large Databases on 64 Bit Workstations

この論文をさがす

抄録

Since the computer applications become more complex, and the users become more sophisticated based on 32 bit environment, the move to 64 bit environment is inevitable. The gains from 64 bit computing environment can be summed up in three aspects: performance, precision and capacity, which should be reconsidered in the applications which manipulate multimedia, scientific, data warehousing databases. They are usually huge in the size. For the 64 bit architecture, LP64 is chosen as solution by Open System community considering the portability, interoperability with 32 bit environment. The Sequoia 2000 Project explored the application of emerging database, network, storage, and visualization technologies of Earth science problems, resulting in a "database-centric" metaphor for scientific computation. About Sequoia 2000 benchmark, one problems is that the data size is huge. For example, the national benchmark data is 16 GB/ytes; another problem is how to represent the complicated operation, such as POSTGRES operator "|", "<|>" and "*&*" defined in its queries. INADA is an application platform for building databases with a database language for object management. We integrating ODMG93 C++ binding interface into INADA and call it INADA/ODMG. For the persistent virtual memory used by INADA, UNIX mmap() function is used mapping virtual memory from file. The current 32 bit INADA shows incapabilities while dealing with above problems, such as 2 GBytes file size limit, since the largest virtual memory available to the users is 2 GBytes, the file larger than that size cannot be mapped. This means that very large database such as Sequoia 2000 cannot be handled by the current 32 bit INADA. The key features of our design is to make use of the gains from the 64 bit architecture, to setup 64 bit INADA/ODMG based our current 32 bit INADA/ODMG, and realize the Sequoia 2000 benchmark on it trying to overcome above problems.

収録刊行物

詳細情報 詳細情報について

問題の指摘

ページトップへ